Well yeah, it potentially gets a bit more complicated than had occurred to me that it might... sorry about that... though I don't think anyone's played enough for it to be too hard yet.
You can do a 'custom games' search for the last x games, so as long as you know when someone finishes a batch of 15 and they don't play any more that day, then you're ok.
So for MereNovice for example, searching now, you can search 'custom date' from 1-12 March excluding double or nothings, and see he's played 30 games exactly at 23%. Then you do a 'custom games' search for the last 15 games and it gives you an ROI of 57% for that batch.
Sharkscope doesn't actually give you a way of finding the ROI of a previous batch though. So if he'd played four batches and you wanted to know the ROI of the second batch, that would be tricky, unless you knew the dates when he'd played them and they hadn't overlapped in with any other batches. That is, if he played 15 games from 1-3 March and then 15 from 4-6 and then 15 from 7-9, it's easy enough because you can do a custom date search on these dates. But if he played 20 from 1-3 and 10 from 4-6 then it's a bit harder to separate out.
If it comes to that then I'll do them for you if you like as penance for not thinking of that in the first place... Obviously you can work it out though with some 'custom game' searches (say someone's played 62 and you want to know the ROI of the second batch, and didn't know the dates of that batch, then you'd have to find the ROI from the last 47 and the ROI from the last 32, and work out what the ROI of the second batch of 15 would have to be to make the difference between the two numbers). In MereNovice's case above, for example, you can work out the ROI of his first batch easily enough at -11%, given the numbers that come out of the two Sharkscope searches.